INDEX

Explanations

terms related to engraved or etched markings, particularly when the value is high

oai_token-act-pair · gpt-3.5-turbo

terms that end with the suffix "ings"

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 9-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.9.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.9.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 earthqu

-0.85

vier

-0.79

isons

-0.72

SIGN

-0.72

WATCH

-0.70

ison

-0.68

ISION

-0.67

ãĥ¯ãĥ³

-0.65

charging

-0.60

apons

-0.60

POSITIVE LOGITS

tons

1.12

hots

1.07

omething

1.01

creen

0.95

pace

0.92

hot

0.92

mith

0.87

cape

0.85

manship

0.85

paces

0.82

Activations Density 0.074%

terms related to engraved or etched markings, particularly when the value is high

terms that end with the suffix "ings"

No Comments

No Known Activations